奇异值能量标准谱在机械振动信号降噪中的应用研究
Application of Singular Value Energy Standard Spectrum in Noise Reduction of Mechanical Vibration Signal

针对机械设备振动信号的宽频带、非线性和非平稳特性，提出了一种基于奇异值能量标准谱的自适应降噪方法。首先利用平均最近邻域发散度判定法计算嵌入维数并对信号进行相空间重构，然后对相空间重构矩阵进行奇异值分解，最后提出奇异值能量标准谱概念，根据能量标准谱拐点位置自适应地确定奇异值的重构阶数以重构有用分量，从而实现信号降噪。数值仿真和柴油机缸盖振动信号分析结果表明，本方法能够有效消除信号中的噪声并保留原信号的特征，提高信噪比。
According to the wide-band, nonlinear and non-stationary characteristics of the mechanical equipment vibration signals, a self-adaptive noise reduction method based on singular value energy standard spectrum was proposed. Firstly, the embedding dimension of phase space reconstruction was calculated by using the method of the average divergence degree of nearest neighbors, and then the phase space reconstruction matrix was decomposed with singular value decomposition. Finally the concept of singular value energy standard spectrum was put forward, and reconstruction order number was determined according to the inflexion position in order to refactor useful component parts, so as to realize the signal noise reduction. The analysis results of numerical simulation and cylinder head vibration signal showed that the proposed method can not only eliminate noise in the signal effectively but also retain the original signal characteristic effectively and enhance the signal-to-noise ratio